machine hangs - SLES9/NFS

[Shylendra Bhat]

Hello,

I am looking answers for the following questions.

Is nfs file lock acquired by client, persistent across the nfs server reboot?
I know that this feature was not there in NFSv3. Does NFSv4 supports this?

Following is the exercise which I did to test nfs file lock.

I have two machines. One among them is the NFS server running on SLES9(kernel 2.6.5-7.79). The other machine mounts the NFS exported filesystem.

I have written a application which locks the files over this nfs mount.

After acquiring the lock, I am restarting the nfs service using "rcnfsserver restart" command.

After the nfs service restart, the client fails to release the lock and is in a hung state. If the mount directory is listed, it shows

"bash: cd: /export: Stale NFS file handle"

After this behavior of the client, server stops responding. Only way to bring back the machine is reboot.

Any help is really appreciated.

Thanks & regards
Shylendra

[Ragu_3]

Probably the server stopped due to excessive CPU load. Can you SSH into the `hung server' and a do a top on that? Kill the runaway process cleanly using SIGKILL. There might not be a need to reboot, start the killed server process again.